Mike, I guess you're using the D2XX library, right?
If so, probably the "FT_Write()" is the function you're using to write to the ports.
One idea probably worth trying is to call this function from different PC application threads using 4K data writes. Just try to synchronize the packet writes between the threads in order to kind of start around the same moment in time. The FTDI library might have an intrinsic limitation with the >2K packets(taking a longer time to return from the function if the data length is above 2K), however calling this function from 2 different threads around the same time, should ideally submit different requests to the driver side and in turn the driver should submit interleaved IRPs to the USB host driver.
And an USB analyzer might help as well here. HW-based ones are best but the SW ones might do the job in this case as well.
Here are a few options for the SW ones:
http://desowin.org/usbpcap/http://www.usblyzer.com/Good luck!